Oh right... Rebooting might be needed for completing the CUDA installation, I forgot...
Now you should be able to install pyDEME too, since the installation environment issue is resolved. Thank you, Ruochun On Monday, May 13, 2024 at 4:17:01 PM UTC+8 [email protected] wrote: > Hi Ruochun, > After the reboot of my computer, the demo runs successfully. Thank you for > your careful help! > > On Friday, May 10, 2024 at 3:06:30 PM UTC+8 Ruochun Zhang wrote: > >> Yes. This error was encountered during a *cudaGetDeviceCount *call, >> which is the first GPU-related call when constructing a DEME simulation >> object. So CUDA failed to get the number of devices. I don't know if >> you've been having success running GPU-based applications on this machine >> but again, I suspect it can't run anything GPU-related, not limited to >> DEME. Possible treatments include... >> 1. The NVIDIA driver installed on your system is not compatible with CUDA >> 12.0, or is corrupted. You probably want to ensure a correct installation >> of a GPU driver first, if you haven't paid attention to doing that. >> 2. You may want to check using *nvidia-smi* to see if the OS recognizes >> the GPU at all. >> 3. You should ensure that you have the permissions needed to access GPUs. >> >> In short, ensure you have GPU access. It's still good to start from a >> minimal CUDA example to debug, since it appears to be a basic environmental >> issue. >> >> Thank you, >> Ruochun >> >> On Friday, May 10, 2024 at 9:11:52 AM UTC+8 [email protected] wrote: >> >>> Hi Ruochun, >>> I have succefully configured the DEME after the reinstallation of CUDA. >>> Thank you! It is indeed that the multiple times installation of CUDA >>> induced the previours error. >>> When I run the demo, there is an error like this. Do you have any >>> suggestion? >>> [image: 微信图片_20240510091101.jpg] >>> >>> On Thursday, May 9, 2024 at 7:08:28 PM UTC+8 Ruochun Zhang wrote: >>> >>>> To me, because CMakeCUDACompilerId.cu is a Cmake-generated minimal >>>> example and it failed to compile, it looks like the compiler is not >>>> configured properly and cannot compile anything. The solution is probably >>>> case-by-case. I am not familiar with the installation process of CUDA >>>> enough to accurately remote-diagnose (but someone else here might be). The >>>> best you can do is to write a minimal CUDA example, test if it can be >>>> compiled, and debug starting from there. However, I have a couple of >>>> guesses... >>>> 1. If you installed CUDA multiple times or had multiple trials (like >>>> once from apt-get and then another time from a downloaded file), there can >>>> be conflicts. The nvcc in /usr/bin might be a symbolic link, and it could >>>> be linked against an executable that is not compatible with the CUDA >>>> toolkit. You might try to fix this symbolic link by linking it against >>>> that >>>> one in cuda-12.0/bin, or the one you think is correct. >>>> 2. Just remove all installation of CUDA and start over. >>>> 3. I don't know the GCC or Clang version you are using. It's not >>>> likely, but it might be too old or too new to be compatible with nvcc. GCC >>>> 9.4 should be a version that works in most of my trials, in case you wish >>>> to give it a try. >>>> >>>> That is all I can think of at this point, thank you, >>>> Ruochun >>>> On Thursday, May 9, 2024 at 4:54:34 PM UTC+8 [email protected] wrote: >>>> >>>>> After press c to configure, there is also an error like this: >>>>> [image: 5c8b56934e5f50fa728c05edc5b5ffd.jpg] >>>>> >>>>> On Thursday, May 9, 2024 at 2:27:59 PM UTC+8 Ruochun Zhang wrote: >>>>> >>>>>> Can you please press c to configure the project, so we might know >>>>>> what is still missing for the project? If nothing goes wrong, then try >>>>>> pressing g to generate the project. >>>>>> >>>>>> Thank you, >>>>>> Ruochun >>>>>> >>>>>> On Thursday, May 9, 2024 at 10:11:05 AM UTC+8 [email protected] >>>>>> wrote: >>>>>> >>>>>>> I updated the cmake from 3.20 to 3.29, which made it find the nvcc. >>>>>>> However, the ccmake GUI is empty now. >>>>>>> [image: 微信图片_20240509100541.jpg] >>>>>>> >>>>>>> >>>>>>> On Thursday, May 9, 2024 at 3:36:40 AM UTC+8 Ruochun Zhang wrote: >>>>>>> >>>>>>> It looks like it found the compiler nvcc this time, but it's not >>>>>>> working correctly... How did you make it find nvcc in */usr/bin* >>>>>>> this time? I thought you had nvcc in another location last time. You >>>>>>> re-installed? And can you show the ccmake GUI this time as well? >>>>>>> >>>>>>> Ruochun >>>>>>> >>>>>>> On Wednesday, May 8, 2024 at 5:16:58 PM UTC+8 [email protected] >>>>>>> wrote: >>>>>>> >>>>>>> Hi Dr. Zhang, >>>>>>> >>>>>>> I still encountered a error like this. Do you have any suggestion? >>>>>>> >>>>>>> Thanks! >>>>>>> [image: a391d46ae4497f81bd18b7e111b4842.jpg] >>>>>>> On Tuesday, May 7, 2024 at 8:20:57 PM UTC+8 Ruochun Zhang wrote: >>>>>>> >>>>>>> Doesn't look like it finds a CUDA compiler at all. It appears that >>>>>>> /usr/bin is checked but it's not found there. Please make sure your >>>>>>> CUDA >>>>>>> installation is in PATH. Maybe the troubleshooting part of this >>>>>>> section >>>>>>> <https://github.com/projectchrono/DEM-Engine?tab=readme-ov-file#compilation> >>>>>>> >>>>>>> of DEME manual can help you. >>>>>>> >>>>>>> Thank you, >>>>>>> Ruochun >>>>>>> >>>>>>> On Tuesday, May 7, 2024 at 7:14:18 PM UTC+8 [email protected] >>>>>>> wrote: >>>>>>> >>>>>>> The ccmake GUI and the configure error are shown below. I do not see >>>>>>> anything related to CUDA_ARCHITECTURES or CUDA_PATH. >>>>>>> [image: 微信图片_20240507191053.jpg] [image: 微信图片_20240507191047.jpg] >>>>>>> >>>>>>> On Tuesday, May 7, 2024 at 6:12:11 PM UTC+8 Ruochun Zhang wrote: >>>>>>> >>>>>>> Looks all good. Please show your ccmake GUI when configuring the >>>>>>> project: You may see in there a few missing items that the >>>>>>> autodetection >>>>>>> failed to identify. Could be CUDA_ARCHITECTURES or CUDA_PATH related. >>>>>>> If this is the case, then we may help you fill them out manually. >>>>>>> >>>>>>> Ruochun >>>>>>> On Tuesday, May 7, 2024 at 5:23:14 PM UTC+8 [email protected] >>>>>>> wrote: >>>>>>> >>>>>>> Hi Dr. Zhang: >>>>>>> >>>>>>> The versions of cmake and cuda, and the model of GPU are shown below. >>>>>>> >>>>>>> Thank you! >>>>>>> [image: 微信图片_20240507171644.jpg] >>>>>>> >>>>>>> On Tuesday, May 7, 2024 at 4:50:27 PM UTC+8 Ruochun Zhang wrote: >>>>>>> >>>>>>> Hi Dr. Shen, >>>>>>> >>>>>>> A couple of questions: What is your cmake version, what is your CUDA >>>>>>> version, and what model is the GPU you have on your system? >>>>>>> >>>>>>> Thank you, >>>>>>> Ruochun >>>>>>> >>>>>>> On Tuesday, May 7, 2024 at 8:37:47 AM UTC+8 [email protected] >>>>>>> wrote: >>>>>>> >>>>>>> Hello everyone, >>>>>>> >>>>>>> When i build the c++ DEME, there is an error as shown in the figure >>>>>>> below. >>>>>>> >>>>>>> Coould anyone give me a help? >>>>>>> >>>>>>> Thanks![image: 微信图片_20240507083122.jpg] >>>>>>> >>>>>>> -- You received this message because you are subscribed to the Google Groups "ProjectChrono" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/projectchrono/76e369f8-61ed-4992-a541-f06f92d75435n%40googlegroups.com.
